Research On Power Transformer Status Diagnosis Technology Based On Bacterial Foraging Algorithms

Power transformer is one of the most important equipments in the power grid.Its performance is closely related to the safe and economical operation of the power system.Therefore,grasping the operation status of power transformer accurately,detecting potential faults and accurate fault location promptly,can effectively reduce the incidence of power accidents and ensure the reliability and safety of power supply.In the current power transformer fault diagnosis and fault location technology,the state judgment standards of the traditional dissolved gas analysis method(DGA)in oil and electrical test are too absolute,therefore,many intelligent algorithms have been introduced into the field of power transformer fault diagnosis,and better results have been gained.However,the accuracy of intelligent algorithm is affected by parameters.Therefore,based on the excellent parameter optimization ability of bacterial foraging algorithm(BFA),this paper studied the condition diagnosis and fault location technology of power transformer..Firstly,this paper studied the characteristics,principle,algorithm model and algorithm steps of bacterial foraging algorithm.The chemotaxis,replication and migration of primitive bacterial foraging algorithm were introduced,The step size was improved to avoid the problems of slow convergence speed and crossing over the optimal solution caused by the fixed step size of traditional bacterial foraging algorithm.Bacterial foraging algorithm was compared with particle swarm optimization(PSO)and genetic algorithm(GA),and both their advantages and disadvantages were analyzed.Secondly,the fault mechanism and fault diagnosis technology of power transformer were studied,and a fault diagnosis optimization model of power transformer based on bacterial foraging algorithm was established.The model took the relative value of characteristic gas content in power transformer oil as the state evaluation sample,took kfold average classification accuracy as objective function,the global optimal parameter solution of support vector machine was searched by bacterial foraging algorithm.The simulation results showed that the bacterial foraging algorithm can select the optimal parameters of support vector machine more quickly than genetic algorithm and particle swarm optimization,and the optimized model had higher accuracy.The result that the fault diagnosis model of support vector machine power transformer based on optimization method of bacterial foraging algorithm can accurately diagnose the data that can not be judged by IEC three-ratio method was showed too.And the validity of the model was verified by an example.Finally,in the research of power transformer fault location technology,the oil chromatographic information of transformer was combined with electrical test characteristics.Eleven characteristic attribute variables of fault location were proposed,and according to the attribute variables,eight fault locations of power transformers were determined.Based on these,a fault location model was established in this paper.The bacterial foraging algorithm was used as a computing tool to cluster the fault location.And according to the calculation results and the principle of maximum membership degree,a complete binary tree model was established.Seven support vector machines were used to classify the binary tree layer by layer and the parameters of the model were optimized based on the bacterial foraging algorithm.A power transformer fault location model based on bacterial foraging algorithm and complete binary tree was established.Compared with other intelligent algorithms,the fault location model can judge the fault location quickly,and had a higher location accuracy were proved by example verification.